Japan Opens a Semipermanent Adaptive Trial Platform for Sepsis
Sepsis research in Japan is moving into a different operational model. The EURASSIA trial is designed as an adaptive platform study, a structure built to test multiple interventions inside a single long-running framework rather than launching one isolated randomized trial after another. In a syndrome as heterogeneous as sepsis, that matters because a treatment signal can be diluted when patients with very different inflammatory states, infection sources, organ failures, and hemodynamic profiles are forced into one broad bucket.
The protocol centers on adults with sepsis who require intensive care across 11 tertiary care centers in Japan. Patients for whom care has been withheld or withdrawn at the time of sepsis diagnosis are excluded, which keeps the trial focused on individuals for whom active treatment remains clinically relevant. The platform began in 2024 and is built to accept new domains over time while others are removed, allowing the study structure to evolve as clinical questions change.
This is not a static trial with a single question and a single answer. It is a semipermanent research engine.
Two Domains Enter the Platform at Launch
At inception, the trial includes two domains running in parallel. Each domain examines a separate management question, and participants are assessed for eligibility within each domain before randomization. One domain evaluates intravenous low-dose landiolol versus no medication for mild tachycardia in the hemodynamic management setting. The other compares chlorhexidine-containing dressing with usual dressing at a catheter insertion site in the infection control setting.
The design reflects a practical response to the biological complexity of sepsis. A patient in septic shock may need cardiovascular rate control, infection prevention, organ support, or all of the above, yet the relative importance of each intervention changes across time and across patient subtype. By separating questions into domains, investigators can examine interventions without assuming that all septic patients respond in the same way to the same maneuver.
In hemodynamic management, low-dose landiolol is a plausible candidate because short-acting beta-blockade may temper excessive tachycardia without locking clinicians into prolonged cardiovascular suppression. In the catheter domain, chlorhexidine-containing dressings target a different mechanism entirely: the local reduction of catheter colonization and subsequent bloodstream infection risk. The two domains therefore test distinct biological pathways inside the same ICU population.
90-Day Survival Anchors the Primary End Point
The trial’s primary outcome is 90-day survival after intensive care initiation for sepsis. That endpoint is clinically substantial because it captures more than immediate resuscitation success. Patients who survive the first hours of sepsis may still die later from progressive organ dysfunction, secondary infection, or complications that emerge after the acute inflammatory surge has passed. A 90-day horizon is long enough to reflect that downstream burden while still remaining tied to the initial critical illness.
Additional outcomes will be defined at the domain level when each domain is opened. This structure gives investigators room to tailor the measurable effects of each intervention to the question being asked. A hemodynamic intervention does not necessarily need the same secondary endpoints as a catheter-care intervention, and the protocol makes room for that difference rather than forcing a single universal template.
Intervention-by-subtype interactions will be examined at interim and final analyses. In sepsis, subtype effects are not a statistical nicety; they are often the reason a treatment appears weak in an unselected cohort and meaningful in a biologically narrower one. The protocol allows those interactions to inform sample size modifications and, when needed, the addition of a stratification factor for random allocation.
Adaptive Randomization Without a Response-Adaptive Scheme
Although the study is adaptive, it does not use reweighted response-adaptive randomization. Instead, the platform relies on a predefined statistical analysis plan within each domain. That choice keeps the framework flexible without letting allocation probabilities drift in ways that can complicate interpretation or operational stability across a busy multicenter ICU network.
The decision also fits the setting. Sepsis trials are exposed to rapidly changing clinical circumstances, but the patient population remains highly vulnerable to noise from site variation, fluctuating severity, and co-interventions. A clear protocolized approach to analysis is often more defensible than a more aggressive response-adaptive design when the goal is to preserve interpretability across multiple domains and centers.
Each domain will carry its own predefined analysis plan, which is essential in a semipermanent platform. Without that discipline, a trial that begins as a single ICU study can become a moving target. Here, the adaptive framework is not a shortcut around rigor; it is the mechanism intended to make rigorous evidence generation feasible in a condition where conventional large randomized trials are slow and expensive.
Why a Platform Trial Fits Sepsis Biology
Sepsis remains one of the most difficult syndromes in critical care because it is both common and biologically uneven. Some patients present with overwhelming vasodilation and tachycardia, others with catheter-related complications, and others with a dominant organ failure pattern that blurs the boundary between infection, host response, and iatrogenic risk. Traditional randomized controlled trials often struggle in this setting because a modest treatment effect can vanish when enrolled patients are too dissimilar for the intervention to work uniformly.
A platform trial addresses that problem by allowing multiple questions to be studied within the same infrastructure. Investigators do not have to rebuild the trial each time a new strategy becomes relevant. Domains can be added as the science evolves, then removed when the question has been answered or no longer warrants further testing. For a disease like sepsis, where treatment pathways are likely to differ by phenotype and clinical context, that efficiency is more than administrative convenience. It is a way to keep pace with the biology of the syndrome.
The protocol also signals a willingness to test both pharmacologic and nonpharmacologic strategies under one governance structure. That breadth matters because sepsis management is not only about choosing a drug. It also includes preventing line-associated complications, controlling infection sources, and adjusting hemodynamics in a way that supports perfusion without worsening cardiac stress.
Ethics Approval and Trial Registration Set the Framework
The Certified Review Board of Keio at the principal institution approved the study under approval number N20240002. Written informed consent will be obtained from participants or their legal representatives, which is essential in critical care research where many patients cannot consent for themselves at presentation. The trial is registered under jRCT1031240424.
Results will be disseminated through publications and presentations once available. For now, the protocol establishes the infrastructure for a multicenter, semipermanent sepsis platform in Japan. The real test will come when the first domains mature, interim analyses begin to examine subtype interactions, and the platform’s ability to accommodate new questions is judged against the practical realities of ICU enrollment, severity stratification, and endpoint clarity.
What remains unresolved is not whether sepsis needs better evidence. It does. The open question is how many clinically distinct questions this platform can answer before the heterogeneity of the syndrome forces another redesign of the trial itself.